Abstract
In December 1999, IBM Research announced a five-year, $100M research project, code named Blue Gene, to build a petaflop computer which will be used primarily for research in computational biology. This computer will be 100 times faster than the current fastest supercomputer, ASCI White. The Blue Gene project has the potential to revolutionize research in high-performance computing and in computational biology. To reach a petaflop, Blue Gene interconnects approximately one million identical and simple processors, each capable of executing at a rate of one gigaflop. Each of the 25 processors on a single chip contains a half megabyte of embedded DRAM, which is shared among those processors. They communicate through a system of high speed orthogonal opposing rings. The approximately 40,000 chips communicate by message passing. The configuration is suitable for highly parallel problems that do not require huge amounts of memory. Such problems can be found in computational biology, high-end visualization, computational fluid dynamics, and other areas. This talk will be primarily about the Blue Gene hardware and system software. We will also briefly discuss the protein folding application.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2000 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Warren, H.S. (2000). Blue Gene. In: Valero, M., Joe, K., Kitsuregawa, M., Tanaka, H. (eds) High Performance Computing. ISHPC 2000. Lecture Notes in Computer Science, vol 1940. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-39999-2_4
Download citation
DOI: https://doi.org/10.1007/3-540-39999-2_4
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-41128-4
Online ISBN: 978-3-540-39999-5
eBook Packages: Springer Book Archive